Assign every residential parcel to a neighborhood
Given that neighborhood structure
Develop trending factors for each neighborhood with more than 3 or 5 sales
Separate trending factors are estimated for land and improvement
Apply the trending factors and evaluate:
The resulting ratio and COD
The percentage of properties trended
Repeat for each of the five years 2013-2017
Repeat for each classification scheme
Current neighborhoods, townships, school districts, census tracts, etc.
Further division based on clustering by acreage, year built, sq. ft., # of rooms
Describe tradeoffs between the different classifications tested.
For an initial pass, we construct a model in which valid sales prices were predicted by land and improvement values. The values taken from this baseline model were then used to calculate updated assessed values.
| COD | 12.986829 |
| Ratio | 1.008983 |
With a very basic model, we were able to achieve reasonable values on both COD and Ratio.
As a next step, we computed individual regressions for each pre-existing neighborhood with at least 5 sales.
Min = 5
| trueCOD | 11.2130345 |
| ratio | 1.0040836 |
| proportion | 0.7210493 |
Min = 3
| trueCOD | 11.0237743 |
| ratio | 1.0041818 |
| proportion | 0.7924216 |
Min = 5
| trueCOD | 12.966301 |
| ratio | 1.004886 |
| proportion | 1.000000 |
Min = 3
| trueCOD | 12.966301 |
| ratio | 1.004886 |
| proportion | 1.000000 |
Min = 5
| trueCOD | 13.4511233 |
| ratio | 1.0225523 |
| proportion | 0.9819206 |
Min = 3
| trueCOD | 13.4542618 |
| ratio | 1.0224384 |
| proportion | 0.9884329 |
Min = 5
| trueCOD | 13.6441386 |
| ratio | 1.0204829 |
| proportion | 0.9972428 |
Min = 3
| trueCOD | 13.6441386 |
| ratio | 1.0204829 |
| proportion | 0.9972428 |
In an attempt to find an additional way to create neighborhoods, we also used Census Tracts.
Min = 5
| trueCOD | 12.6603262 |
| ratio | 1.0092367 |
| proportion | 0.7743947 |
Min = 3
| trueCOD | 12.6464636 |
| ratio | 1.0092221 |
| proportion | 0.7866446 |
After examining individual geographical areas, we made an effort to nest geographical features into two levels.
| trueCOD | 13.6027716 |
| ratio | 1.0198828 |
| proportion | 0.9972428 |
| trueCOD | 13.4511233 |
| ratio | 1.0225523 |
| proportion | 0.9819206 |
| trueCOD | 12.5715277 |
| ratio | 1.0098171 |
| proportion | 0.8331758 |
Given the relative performance of tract, an effort was made to break the homes within tract into discreet clusters. The features selected clustering were acreage, year built, finished dwelling area and finished rooms. For each individual school district or tract, 2 distinct clusters were created.
| trueCOD | 12.9986636 |
| ratio | 1.0042643 |
| proportion | 0.9913214 |
| trueCOD | 12.0229593 |
| ratio | 1.0083396 |
| proportion | 0.7122525 |
Instead of fitting models to individual groupings, we also tried a mixed model in which every tract X cluster combination has its own slope within the model.
| COD | 10.597028 |
| Ratio | 1.009745 |
Given the reasonable results of the tract X cluster combination, we can see how it performs over the previous years.
| trueCOD | 14.3687358 |
| ratio | 1.0417522 |
| proportion | 0.9972347 |
| trueCOD | 10.7265009 |
| ratio | 1.0021303 |
| proportion | 0.6956888 |
| trueCOD | 12.3242613 |
| ratio | 1.0045554 |
| proportion | 0.7119249 |
| trueCOD | 12.8624315 |
| ratio | 1.0187185 |
| proportion | 0.9972983 |
| trueCOD | 10.2757028 |
| ratio | 1.0024316 |
| proportion | 0.6486864 |
| trueCOD | 11.8786006 |
| ratio | 1.0045146 |
| proportion | 0.6678112 |
| trueCOD | 12.0910325 |
| ratio | 1.0269729 |
| proportion | 0.9972973 |
| trueCOD | 9.092055 |
| ratio | 1.001457 |
| proportion | 0.629402 |
| trueCOD | 10.4149531 |
| ratio | 1.0034560 |
| proportion | 0.6213207 |
| trueCOD | 16.8158823 |
| ratio | 1.0283018 |
| proportion | 0.9910053 |
| trueCOD | 10.7369317 |
| ratio | 0.9971802 |
| proportion | 0.5721526 |
| trueCOD | 12.0105377 |
| ratio | 0.9939950 |
| proportion | 0.6179198 |